RadShield - Radiation Proctitis Risk Evaluation

A modern, medical-grade web application designed for Radiation Oncologists to assess, visualize, and monitor the risk of Radiation Proctitis (RadPro) in patients undergoing radiation therapy.

This project upgrades a legacy Streamlit prototype into a full-featured, responsive React application integrated with Google's Gemini 2.5 Flash AI for intelligent clinical decision support.

---

📸 Interface Preview

1. Department Dashboard

A high-level overview of department statistics, recent critical cases, and risk distribution.

2. AI Risk Assessment Form

Structured clinical intake form integrating demographics, dosimetry, and AI-analyzed clinical notes.

3. Analysis Results & Clinical Decision Support

Comprehensive risk report featuring multi-dimensional radar charts, population positioning, and AI-generated reasoning.

4. ePRO System (Patient View)

Mobile-responsive interface for patients to report symptoms and complete standardized questionnaires (QLQ-PRT20).

5. Gemini Medical Assistant

Integrated chatbot for real-time medical queries and data interpretation.

---

🌟 Key Features

1. Intelligent Risk Assessment

• AI-Powered Analysis: Utilizes Google Gemini 2.5 Flash to analyze structured data (Dosimetry, Demographics) and unstructured text (Clinical Notes, Genomic Markers).
• Dual-Risk Scoring: Calculates distinct probability scores for Acute Toxicity (0-3 months) and Late Toxicity (>6 months).
• Multi-Dimensional Radar: Visualizes risk contributions across 5 key axes: Genomics, Dosimetry, Age, Medications, and Comorbidities.

2. Professional Medical UI

• Modern Dashboard: A clean, "Glassmorphism" inspired interface designed for clinical environments.
• Interactive Visualizations:
  • Radial Gauges: Instant visual feedback on risk levels.
  • Population Bell Curve: Places individual patient risk in the context of a 1,200-patient dataset.
  • Animated Radar Charts: Clear visualization of multi-factorial risk drivers.
• Responsive Design: Built with Tailwind CSS for seamless use on desktops and tablets.

3. ePRO System (Patient Reported Outcomes)

• Integrated Patient Portal: A dedicated interface for patients to report symptoms directly.
• Standardized Questionnaires: Fully integrated EORTC QLQ-PRT20 (European Organization for Research and Treatment of Cancer) module for standardized quality-of-life assessment.
• Symptom Scoring: Automatically calculates normalized Symptom Burden Scores (0-100) based on patient responses to the 21-item questionnaire.
• General Symptom Tracking: Tracks Rectal Bleeding status, Pain Index (0-10), Urgency, and Stool Frequency.
• Longitudinal History: Visualizes trends for both general symptoms and standardized QLQ-PRT20 scores to help clinicians identify progression early.

4. Clinical Decision Support (CDS)

• Automated Reasoning: The AI generates human-readable reasoning for every risk score.
• Actionable Recommendations: Provides tailored clinical next steps (e.g., "Review anticoagulant therapy", "Schedule colonoscopy").
• Contraindication Checks: Automatically flags potential contraindications based on patient history.

---

🛠️ Technology Stack

• Frontend: React 19, TypeScript, Vite
• Styling: Tailwind CSS (Custom "Medical" color palette & Animations)
• AI Engine: Google GenAI SDK (@google/genai) connecting to Gemini 2.5 Flash
• State Management: React Hooks (useState, useEffect, useRef)
• Icons: Custom SVG Icons tailored for medical context

---

🚀 Getting Started

Prerequisites

• Node.js (v18 or higher)
• A Google AI Studio API Key (for Gemini integration)

Installation

1. Clone the repository

   git clone https://github.com/your-username/radshield-web-ui.git
   cd radshield-web-ui

2. Install Dependencies

   npm install

3. Configure API Key

   • Create a .env file in the root directory.
   • Add your Gemini API Key:

     API_KEY=your_google_ai_studio_api_key_here

   • Note: In the current prototype setup via standard build tools, ensure the key is accessible to the process environment.

4. Run Development Server

   npm start

   Open http://localhost:8080 to view the app.

---

📂 Project Structure

• App.tsx: Main application logic, routing (Dashboard/Assessment/History/ePRO), and view components.
• services.ts: Handles interaction with the Google Gemini API, including prompt engineering and JSON schema validation.
• types.ts: TypeScript definitions for Patient Data, Risk Models, and ePRO records.
• constants.ts: Mock data for history, dashboard statistics, questionnaires, and configuration options.
• index.html: Root HTML file containing Tailwind CSS CDN and font configurations.

---

🩺 Clinical Context

Radiation Proctitis is a common side effect of pelvic radiation therapy (e.g., for prostate or cervical cancer). Early identification of high-risk patients allows for:

1. Dose Optimization: Adjusting VMAT/IMRT plans for better sparing.
2. Pre-habilitation: Managing comorbidities (diabetes, hypertension) before treatment.
3. Early Intervention: Initiating symptom management protocols immediately upon report.

RadShield bridges the gap between complex dosimetric data and actionable clinical insight.

---

⚠️ Disclaimer

This application is a prototype for research and demonstration purposes. It is not an FDA-approved medical device. The risk scores and AI recommendations should be verified by a qualified medical professional before making clinical decisions.